Public-Private-Sector Employment Decisions and Wage Differentials in Peninsular Malaysia

Transcription

1 Public-Private-Sector Employment Decisions and Wage Differentials in Peninsular Malaysia Ganesh K. Seshan ABSTRACT: This paper examines whether there is a wage premium for public-sector workers in Peninsular Malaysia and whether there is wage discrimination based on gender. Public- and private-sector wages are estimated using individual-level data from 1995 and 2007 while accounting for sectoral choice by men and women. I find that public-sector employees earn a wage premium, independent of their human capital endowments and personal attributes. This wage premium has increased over the twelve-year period. There is little evidence of a gender wage gap in the public sector; a gender wage gap is more evident in the private sector. KEY WORDS: gender gap, Malaysia, public-private wage differential. A growing number of studies look at the difference in public- and private-sector wages in developing countries (Assaad 1997; Nielsen and Rosholm 2001; Tansel 2005; Terrell 1993), where the public sector usually makes up a major share of wage employment. However, to date, little has been written on public-private wage differentials in Southeast Asia. 1 This study examines the public-private wage differential in Peninsular Malaysia using recently available household data from 2007 and contrasts it with outcomes from The aim of this study is to determine whether public-sector workers in Malaysia earn a wage premium above the private-sector wage. The analysis is further nuanced by looking at gender dimension in wage differentials. Why does a study of the public-private wage differential matter? In most Southeast Asian countries, the private sector, as opposed to the public sector, is a key driver of economic growth. If wages and other benefits are higher in the public sector relative to those offered in the private sector, individuals, even those with skills demanded by the private sector, may queue for employment in the public sector. This would compromise the ability of the private sector to attract the necessary human capital to fill vacancies, improve productivity, and foster innovation. A higher public-sector wage bill resulting from the existence of such wage premiums also places a burden on government expenditures and may divert resources from other critical areas. Nonetheless, arguments could be made in favor of a wage premium in the public sector. Such arguments include reducing corruption amongst civil servants and enhancing productivity. Figure 1 is a useful starting point to examine the public-private wage differential in Peninsular Malaysia using multiple year observations of annual wage earnings data from the Malaysia Household Income Surveys (MHIS) from 1995 to The data show the trend in average annual wage earnings, expressed in 2000 Malaysian ringgit (MYR) to remove the effect of inflation, for public-sector and private-sector employees and include gross domestic product (GDP) per capita as a comparison. 3 For the purpose of this study, the public sector in Malaysia consists of public administration and defense, education (excluding self-employed individuals), and the utilities sector. The utilities sector is Ganesh K. Seshan (gks7@georgetown.edu) is an assistant professor of economics at the Georgetown University School of Foreign Service in Qatar, Doha, Qatar. Emerging Markets Finance & Trade / November December 2013, Vol. 49, Supplement 5, pp M.E. Sharpe, Inc. All rights reserved. Permissions: ISSN X (print)/issn (online) DOI: /REE X4905S510

2 164 Emerging Markets Finance & Trade Figure 1. Malaysia: real annual public- and private-sector wage earnings, Source: Author s calculations using MHIS, multiple years. GDP per capita data from World Development Indicators. Note: Amounts are expressed in 2000 Malaysian ringgit (MYR). composed primarily of government-linked companies (GLCs) that are partially or wholly owned by the federal government. In 2007, the public sector employed 13.9 percent of the total workforce for the states covered in the MHIS data set. 4 The formal private sector accounts for the largest share of employment in Malaysia, 5 and this is reflected in the close comovement of private-sector wages and GDP per capita. Over the twelve-year period, inflation-adjusted public-sector wages grew by 4.26 percent annually while real wage earnings in the private sector rose at a rate of 3.00 percent per year. The result is a widening gap between the average annual wages received in the public sector and those received in the private sector. In 1995, the average ratio of public to private wages was 1.12; by 2007, it had grown to Thus, between 1995 and 2007, the average publicsector wage increased from being 12 percent greater than the average wage in the private sector to being 29 percent higher at the end of the twelve-year period. It would be misleading to rely on a simple average of the wage differential between public and private employees to determine the presence of a public-sector wage premium without accounting for differences in human capital endowment, personal and household characteristics, and local market conditions that influence wage earnings. I investigate how individuals are selected into public-sector employment and (formal) private-sector work and examine the wage determinants in both sectors by men and women separately using MHIS data from 2007, the year for which the public-private wage differential appears to be the largest over the prior twelve-year period. Using a four-way multinomial logit model, I distinguish among nonparticipation, public-sector work, private-sector work, and other employment. Mincerian wage equations are estimated for the public and private sector taking into account sector selection followed by Oaxaca Blinder decomposition of the wage differential between sectors by gender and between male and female by sector. The results show that public-sector employees earn a wage premium over their privatesector counterparts independent of their human capital endowments. This unexplained wage premium, in fact, increased between 1995 and Starting wages are higher in the public sector, and pay increments are based more on seniority than on reward for greater productivity. The public-sector wage premium is relatively greater for female employees than for male workers. A large gender gap is also found, with wages favor-

3 November December 2013, Volume 49, Supplement ing men over women in the private sector; there is greater gender parity in salaries for public-sector employees. Labor Markets in Malaysia A recent report by the World Bank (2010) indicated that labor markets in Malaysia are not sufficiently competitive, and this may provide greater bargaining power to (privatesector) employers than to workers, enabling wages to be kept low relative to the public sector. The report further states that difficulties in creating a new firm in Malaysia suggest that existing firms are not concerned with new competition and could get away with wage offers that do not compensate for productivity since employees have limited access to higher-paying opportunities. A further symptom of noncompetitiveness of the labor markets is that wages have not been growing in tandem with labor productivity. According to a 2009 World Bank report on productivity and investment climate in Malaysia, industry profits in Malaysia are derived from low wages as opposed to innovative products and processes. Low wages could also be a result of low qualifications. Formal schooling in Malaysia begins with six years of primary school followed by three years of lower secondary, where students sit for a Lower Certificate of Education (LCE) or Sijil Rendah Pelajaran (SRP) exam in Form Three. 6 This is followed by another two years of upper secondary where students take the Malaysia Certificate of Education (MCE) or Sijil Pelajaran Malaysia (SPM) examination in Form Five. Preuniversity, which is Form Six, requires two additional years to complete with students taking a Higher School Certificate (HSC) or Sijil Tinggi Persekolahan Malaysia (STPM) examination. A diploma may take up to two years with an undergraduate degree requiring three years after Form Six or four years after Form Five, inclusive of a year of matriculation. Approximately 74 percent of the Malaysian labor force in 2007 had completed upper-secondary level (SPM) or lesser qualification as reflected in the MHIS data. The share of tertiary-educated workers, which was 8 percent in 2007, 7 is still low relative to other competitive East Asian countries such as South Korea and Singapore. Malaysia scores 8.0 for tertiary education in its Human Development Index (HDI) education subcomponent, compared to 23.4 for South Korea and 19.6 for Singapore. The public sector in Malaysia continues to play an important role in the local labor market despite significant privatization efforts since the 1980s. Total public-sector employment increased from 9.8 percent in 1998 to 14.0 percent in the fourth quarter of 2009 (Choong et al. 2010). The total central government wage bill was 26.1 percent of overall expenditure between 1996 and 2000, higher than the average for countries in the East Asia and Pacific region (24.4 percent) and for middle-income countries (24.7 percent) over the same period (United Nations 2005). Econometric Model and Data Econometric Model Studies of public-private wage differentials are based on the model of wage determinations developed by Mincer (1974). Variations in (log) wages are due to differences in human capital characteristics and location factors, which capture labor market conditions and cost-of-living differences. The corresponding wage function takes the form

4 166 Emerging Markets Finance & Trade lnw i = β 0,i + β i X i + µ i, (1) where W denotes annual wages, β is the vector of coefficients associated with wagedetermining attributes X, and µ is a normally distributed random disturbance term with i representing public sector or private sector. Estimating Equation (1) using ordinary least squares (OLS) is problematic, as workers do not randomly choose their sector of employment. If some unobservable worker characteristics that determine wages are correlated with unobserved characteristics that influence the sector of employment, the results will be inconsistent. This is the selection bias problem that Heckman (1979) highlighted. Following an empirical methodology similar to that used by Tansel (2005), to account for sectoral choices, I assume that individuals face four mutually exclusive choices: unemployment, public sector employment, private sector employment, or other employment. The category of other employment includes all other employed individuals who are not salaried employees in the public sector or private sector. This includes the self-employed, employers, and unpaid family workers. Sectoral choice will depend on the worker s preferences, human capital endowment, household attributes, and perceived difference in wage and nonwage compensations across the sectors. I use a conditional multinomial logit model for the probability that an individual chooses among alternative i as follows: 3 Pi = exp ( Zγi) / + ( Z i ) 1 exp γ i, = 1 where γ is a vector of coefficients of the alternative i associated with variables Z affecting sector choice. A two-step estimation method developed by Lee (1983) and Trost and Lee (1984) is adopted. In the first stage, sectoral choice probabilities are estimated using a maximum likelihood logit method, and the selection term is constructed for the alternative i using the formula λ i = φ(l i )/(Φ(L i ), where L i = Φ 1 (P i ), where φ is the standard normal density function and Φ is the standard normal distribution function. In the second stage, the estimated selection term λ i is added to the wage equation for each sector of interest. This allows each sectoral wage regression to be estimated consistently using OLS by taking the form lnw i = β 0,i + β i X i + θ i λ i + µ i, (3) The explanatory variables in the wage regressions include dummy variables for different education levels from completing primary schooling through having a university degree. Work experience is measured by subtracting number of years of schooling from age minus six, the age at school entry. 8 A quadratic term in experience is also added. 9 Dummy variables for districts within states are included to control for differences in labor market conditions and cost of living. The same variables are included in the multinomial logit equation with additional variables to achieve identification. Identification is accomplished by finding exogenous variables that influence the reservation wage or sectoral choice but not the offered wage and therefore can be used to differentiate between the decision to participate in the wage sector and the determinants of earnings. I use three variables for this purpose nonlabor income by other household members, 10 a dummy for the existence of other public-sector workers in the household, and a dummy for having other private-sector workers in the (2)

5 November December 2013, Volume 49, Supplement household. 11 Individuals with large nonemployment income will be less likely to work for wages due to the income effect. The presence of other household members in particular sectors may provide labor market information and even connections that should be positively associated with participation and sectoral choice, a point made by Tansel (2005). Data This analysis is based on the 2007 Malaysia Household Income Survey conducted by the Department of Statistics, Malaysia. The sample is restricted to individuals of Malaysian citizenship ages fourteen to sixty-five years, which yields 55,678 individuals residing in six out of eleven states in Peninsular Malaysia and the Federal Territory of Kuala Lumpur. 12 Table 1 presents characteristics of the wage earners examined for this section. Wages are the sum of cash earnings, bonuses, allowances, and the value of income in kind. Annual public-sector wages for men and women are higher on average than annual formal private-sector wages though this could reflect relatively higher schooling attainment in the former. A means test of difference also finds average annual public-sector wages being significantly higher than average annual private-sector wages for both men and women by MYR 5,019 (USD 1,321) and MYR 7,431 (USD 1,956), respectively, at the 1 percent level of significance. 13 The gender gap in average wage is much narrower in the public sector with females receiving MYR 874 less than men; females in the private sector receive MYR 3,286 less than their male counterparts. Public-sector workers are, on average, older, have more work experience, and are better educated than private-sector employees. Women in both sectors tend to have higher educational attainment than men. Nearly 91 percent of women and 79 percent of men in the public sector have an upper secondary diploma (SPM) or greater, compared to 70 percent of women and 60 percent of men in the private sector. 14 Education measured in years of schooling also supports the observation that public-sector employers are more educated than those in the formal private sector. Estimation Results Labor Market Participation Next, I determine the extent to which factors such as educational attainment, age, household unearned income, and choice of employment by other household members influence the sectoral choice of workers. Participation equations are estimated using a multinomial logit model separately for men and women as shown in Tables 2 and 3, respectively. The results provide the marginal effects of each variable or the probability of entering a particular sector calculated at the mean values of the variables. 15 The probability of employment rises with age in all three sectors of work, though at a decreasing rate, in contrast to nonparticipation. For women, educational attainment at the lower secondary level (SRP) or greater raises the probability of working in the public and private sector but apparently reduces the probability of participating in the other employment category. For males, education at the primary level or higher raises the likelihood of participating in the public sector. However, for men, holding a preuniversity qualification or higher reduces the probability of working in the private sector and other employment categories. Education at the preuniversity level (STPM) or higher is associated with greater probability of public employment for both men and women.

6 168 Emerging Markets Finance & Trade Table 1. Malaysia: descriptive statistics of wage earners in 2007 Public Sector Private Sector Male Female Male Female Mean Deviation Mean Deviation Mean Deviation Mean Deviation Real annual wages 23, , , , , , , , Age Work experience Education Less than primary Primary school Lower secondary Upper secondary Preuniversity Diploma University degree Years of schooling Attended religious school Real unearned household income 5, , , , , , , , Other public sector Other private sector

7 November December 2013, Volume 49, Supplement States Johor Kedah Pinang Perak Perlis Selangor Federal Territory of Kuala Lumpur Number of observations 2,523 1,883 13,276 7,406 At least upper secondary At least preuniversity Source: Author s calculations using MHIS Notes: deviations are not reported for indicator variables. Wages and income are measured in 2000 Malaysian ringgit (MYR), where MYR 3.80 = USD 1.00.

8 170 Emerging Markets Finance & Trade Table 2. Malaysia: multinomial logit estimates of sector of employment choices of men, 2007 Public sector Private sector Other employment Marginal effect deviation Marginal effect deviation Marginal effect deviation Age 0.013*** *** *** Age squared/ *** *** ** Education Less than primary *** Primary school 0.046* *** Lower secondary 0.093*** ** Upper secondary 0.152*** *** Preuniversity 0.263*** *** *** Diploma 0.309*** *** *** University degree 0.372*** ** *** Log unearned household income 0.004*** *** ** Other public employees 0.123*** *** *** Other private-sector employees 0.027*** *** *** Pseudo R Total observations 27,683 Source: Author s calculations using MHIS Notes: All the regressions include dummy variables for districts and for number of children ages six and below. errors are corrected for clustering at the district level. *** Statistical significance at the 1 percent level; ** statistical significance at the 5 percent level; * statistical significance at the 10 percent level.

9 November December 2013, Volume 49, Supplement Table 3. Malaysia: multinomial logit estimates of sector of employment choices of women, 2007 Public sector Private sector Other employment Marginal effect deviation Marginal effect deviation Marginal effect deviation Age 0.008*** *** *** Age squared/ *** *** *** Education Less than primary *** ** Primary school *** * Lower secondary 0.025*** Upper secondary 0.087*** *** Preuniversity 0.322*** Diploma 0.477*** ** *** University degree 0.555*** *** Log unearned household income 0.003*** *** *** Other public employees 0.035*** *** *** Other private-sector employees 0.008*** *** *** Pseudo R Total observations 27,995 Source: Author s calculations using MHIS Notes: All the regressions include dummy variables for districts and for number of children ages six and below. errors are corrected for clustering at the district level for a total of forty-four districts. *** Statistical significance at the 1 percent level; ** statistical significance at the 5 percent level; * statistical significance at the 10 percent level.

10 172 Emerging Markets Finance & Trade Income effects on participation, measured by unearned income by other household members, 16 is statistically significant and negative, with the exception of males in the other employment category, as higher unearned income would reduce the need to seek paid employment. The presence of other household members in the public sector raises the probability of participating in public-sector employment for both men and women. Similar to the findings of Tansel (2005) in her study of Turkish workers, the effect is stronger for men than women. The presence of other household members employed in the private sector increases the likelihood of working in the private sector, with the effect, again, being larger for men than for women. Individuals whose household members are in the public or private sector are less likely to participate in the other employment category. These results suggest the role of connections and labor market information conveyed through household members as determinants of sectoral employment. Wage Equations Selectivity-corrected wage regressions using the Mincer framework are used to calculate the private returns to education and the degree to which other individual and local market characteristics may influence earnings in the public and private sector. Inclusion of the selection term, which is constructed from the earlier participation equation, controls for the fact that workers do not randomly work in either sector and allows us to test whether factors influencing the sorting of workers into these two sectors also affect wage earnings. The independent variables common to each wage equation are work experience, educational attainment indicators, and dummy variables for districts. The resulting estimates for men and women are given in Table 4. Work experience exerts a positive influence on wage offers in all sectors while the quadratic term has the expected negative sign in all sectors. All else being equal, men s wages peak at thirty years of experience in both public and private sectors; women s wages peak at thirty-four and twenty-six years of experience in the public and private sectors, respectively. With the exception of primary schooling for workers in the public sector, where only a small proportion have completed only primary school, the coefficients on educational attainment are statistically significant at the 1 percent level or lower across public- and private-sector employment for men and women. The coefficient on the selection term for female employees in the private sector is positive and statistically significant, which implies that women who choose to work in the private sector have higher productivity than the average worker. I could not reach any conclusion for men in the private sector as the selection term was negative and statistically insignificant. For men and women in the public sector, the selection term is negative and statistically significant, indicating that individuals who select public-sector employment have lower productivity than the average worker. Wage Differential Decompositions Figure 2 revisits the MHIS data on annual salaries in the public and private sectors, differentiating by gender between 1995 and The graphs clearly show a widening gap between public- and private-sector wages over this twelve-year period, with the differential being larger for females than males by Observing wages of men and women by sector, we see that the gender gap is narrower in the public sector than in the private sector.

11 November December 2013, Volume 49, Supplement Table 4. Malaysia: selectivity-corrected wage regressions, 2007 Male Female Public sector Private sector Public sector Private sector Coefficient deviation Coefficient deviation Coefficient deviation Coefficient deviation Work experience 0.066*** *** *** *** Work experience 0.111*** *** *** *** squared/100 Education Primary school *** * Lower secondary 0.264*** *** ** *** Upper secondary 0.596*** *** *** *** Preuniversity 0.802*** *** *** *** Diploma 0.936*** *** *** *** University degree 1.169*** *** *** *** Selection term 0.478*** *** *** Constant 8.754*** *** *** *** R Number of uncensored observations 2,523 13,276 1,883 7,406 Turning points Source: Author s calculations using MHIS Notes: All the regressions include dummy variables for districts across seven states in Peninsular Malaysia. errors are corrected for clustering at the district level for a total of forty-four districts. *** Statistical significance at the 1 percent level; ** statistical significance at the 5 percent level; * statistical significance at the 10 percent level.

12 174 Emerging Markets Finance & Trade (a) Male workers (b) Female workers Figure 2. Malaysia: real annual wages by sector and gender, Source: Author s calculations using multiple years of MHIS. Amounts are expressed in 2000 MYR. As stated in the introduction, using a simple average of the wage differential between public- and private-sector employees or between genders to determine the presence of a wage premium is misleading; doing so fails to account for differences in human capital endowments and other personal or locational characteristics that influence earnings. Column (1) of Table 5 shows the decomposition of public-private wage differentials for men and women using Oaxaca-Blinder decomposition. There are four possible components, including the differential caused by selectivity bias, lnwpub ln Wpvt = ( βpub, 0 βpvt, 0)+ 05. ( Xpub Xpvt )( βpub βpvt ) 1st 2nd ( Xpub + Xpvt )( βpub βpvt )+ θpubλpub θ pvtλ pvt, 3rd ( ) where bar denotes the means of the variables. Subscript pub refers to the public sector; pvt denotes the private sector. The θs are the coefficients of the selection terms in the wage equations; the βs are the remaining coefficients of the wage equations. The first component is the difference in the base wage (constant terms), which is often understood as the pure rent from being in the public sector (Terrell 1993). The second component is due to differences in workers endowments, and the third is due to the differences in the coefficients or market returns to the endowments. The final component is due to the 4th

13 November December 2013, Volume 49, Supplement Table 5. Malaysia: wage differentials by sector and gender, 2007 (1) (2) Public-private wage differentials Male-female wage differentials Men Women Public sector Private sector Observed wage gap (0.05) (0.07) (0.02) (0.03) Components attributed to Constant (0.17) (0.27) (0.32) (0.09) Endowments (0.03) (0.03) (0.01) (0.02) Coefficients (0.11) (0.17) (0.21) (0.06) Selection Total unexplained differential (0.13) (0.17) (0.16) (0.06) Source: Author s calculations using wage regressions from Table 4. Notes: Each component is evaluated at the sample means. Total unexplained differential is the sum of constant term and coefficient component. errors are in parentheses and are calculated using the delta method. difference in the selection terms. The sum of the first and third components is typically referred to as the unexplained differential, which may also include omitted factors. Public-Private Wage Differential The observed wage gap between public- and private-sector employees for men and women is 36.9 percent and 56.2 percent, respectively, with public-sector wages being higher than private-sector wages. 17 More than half the observed wage gap can be explained by differences in endowment. From Table 5 (column 1), we notice the higher educational attainment of public-sector workers compared to private-sector workers and recall that, on average, 19 percent more men and 21 percent more women in the public sector have an upper secondary diploma (SPM) or greater compared to their private-sector counterparts. These endowment differences imply that men and women would receive higher earnings in the public sector when both are paid absent any rents or unexplained premium in the wage structure. After accounting for selection, 29 percent of the wage differential between male public-sector employees and their private-sector counterparts can be explained by higher qualifications of the former. For female workers, 25 percent of the public-private wage differential is accounted for by higher educational attainment of females in the public sector. 18 Therefore, the decomposition results in column (1) of Table 5 reveal that public-sector workers receive a substantial unexplained wage premium. Almost all of this unexplained premium results from differences in the intercepts or constant term with the remainder emanating from differences in market returns to wage-determining characteristics, though

14 176 Emerging Markets Finance & Trade the latter is statistically insignificant at conventional levels. Hence, in Peninsular Malaysia, public-sector employees earn a rent, after controlling for other wage determinants. Tansel (2005) finds a wage premium for public administration and state-owned enterprise workers for Turkey. Terrell (1993) also found that public administration workers earned a sizeable rent. Similarly, Lindauer and Sabot (1983) show that wage premium was the most important determinant of public-private wage differentials in Tanzania. Male-Female Wage Differential I examine the extent of gender discrimination separately in the public and private sectors by doing a similar decomposition exercise using the Oaxaca Blinder method with results in column (2) of Table 5. I put discrimination in quotation marks because it reflects not only labor market discrimination but also wage differences due to other omitted factors. The observed wage gap between men and women is relatively lower in the public sector than in the private sector. One reason the average wage differential between men and women in the public sector is small is the higher human capital endowments and private returns to endowments favoring women. The unexplained differential or gender discrimination term is small and statistically insignificant at conventional levels, which reinforces the view that the government has been able to implement policies that reduce or eliminate wage discrimination. However, I find the observed gender wage differential in the private sector is largely attributed to a male premium (the constant term) and market returns in favor of men. The high total unexplained differential between male and female employees in the private sector is suggestive of wage discrimination against women. 19 Evolution of Wage Differentials ( ) The evolution of wage differentials in Malaysia can be examined by contrasting outcomes over a twelve-year period from 1995 to Using the same empirical methodology applied to MHIS 2007 data, I compute Oaxaca Blinder decompositions for wage differentials with 1995 MHIS data. Table 6 provides wage decomposition for wage differentials between public and private sector and by gender in The public-sector wage premiums for men and women were smaller in 1995 than those found for 2007, and so was the gender gap favoring males in the private sector. Human capital endowments also contributed more to explaining the public-private wage differential in 1995 than in Hence, the widening gaps in average wages between public-sector and private-sector employees between 1995 and 2007, as seen in Figure 1, can be attributed more to rising wage premiums in the public sector and less to differences in human capital endowments across sectors. Conclusion This paper provides new evidence of a wage premium for public-sector workers in Peninsular Malaysia in recent years after accounting for differences in education, experience, and sectoral employment choice. The study finds that the unexplained premium or rents from working in the public sector rose between 1995 and These rents can be regarded as conservative estimates as they do not account for nonwage

15 November December 2013, Volume 49, Supplement Table 6. Malaysia: wage differentials by sector and gender, 1995 (1) (2) Public-private wage differentials Male-female wage differentials Men Women Public sector Private sector Observed wage gap (0.05) (0.05) (0.02) (0.02) Components attributed to Constant (0.11) (0.22) (0.25) (0.08) Endowments (0.02) (0.03) (0.02) (0.01) Coefficients (0.06) (0.13) (0.14) (0.05) Selection Total unexplained differential (0.09) (0.13) (0.14) (0.06) Source: Author s calculations using empirical methodology for obtaining selectivity-corrected wage equations as applied to MHIS data for Notes: Each component is evaluated at the sample means. Total unexplained differential is the sum of constant term and coefficient component. errors are in parentheses and are calculated using the delta method. benefits that are received by public-sector employees, which are usually higher than those found in the private sector. A 2009 World Bank report stated that labor markets are uncompetitive in Malaysia, thereby enabling private-sector employers to keep wages low relative to the public sector. A symptom of this is that private-sector wages have not been growing in tandem with labor productivity. Further work is needed to understand this phenomenon and the extent to which it would require additional regulations to reduce barriers that limit productivity-driven wage growth. This paper also shows that there is greater parity in wages earned between men and women in the public sector suggestive of government wage policies that have been effective in reducing gender wage discrimination. In contrast, a noticeable gender wage gap favoring men is apparent in the private sector. Females in the private sector earn the least at all levels of work experience and education, which is suggestive of the presence of wage discrimination in the private sector. However, the gender wage gap may be overstated to the extent that a relatively higher share of females in the private sector are involved in part-time employment, which could not be accounted for with the existing data set. Better statistics in future household or labor force surveys with respect to the nature of employment or hours worked per week would help address this concern. Though government policies can be credited for narrowing the gender wage gap in Malaysia s public sector, more needs to be done to better align wages between men and women in the private sector. Despite a policy of equal pay for equal work in Malaysia since 1969, its application has fallen short in the private sector, partly due to a lack of

16 178 Emerging Markets Finance & Trade proper mechanisms to implement and monitor the equal pay policy (Fernandez 2006). Additional policies that further support female employment, such as flexible working arrangements, parental leave, and child-care facilities, should also be initiated. Finally, it is imperative to address gender stereotypes about the abilities, preferences, appropriate positions, and societal roles for men and women in the country. This could involve media efforts to increase gender sensitivity and eliminate gender stereotypes. Notes 1. Hou (1993) looked at public-private wages in Taiwan while Danh and Long (2006) examined the wage differentials in Vietnam. The last published study of public- and private-sector wages for Malaysia was undertaken by Mazumdar (1981) with employment data from Data were only made available to the researcher for the Federal Territory of Kuala Lumpur and for six out of eleven states in Peninsular Malaysia, namely Perlis, Kedah, Pulau Pinang, Perak, Selangor, and Johor. The five states that are unaccounted for are Terengganu, Kelantan, Pahang, Negeri Sembilan, and Melaka. 3. Before 2006, an implicit peg was maintained in which USD 1.00 = MYR In 2006, the currency was permitted to float and began to appreciate; by 2007, USD 1.00 = MYR Only Malaysians can be employed in public administration, where the mandatory retirement age is fifty-five. 5. Private-sector employees represented 65 percent of all employed individuals in 2007, up from 59 percent in 1995, based on estimates using the MHIS surveys. 6. The school system was modeled after a British curriculum. 7. The share of workers with degrees was 3.5 percent in 1995, computed using data from the 1995 MHIS. 8. For individuals with religious schooling, work experience was computed as age minus fourteen years of schooling minus six. 9. Though not reported, in part due to the data-sharing agreement, I also ran a separate set of regressions including dummies for ethnicity (Malay, Chinese, and Indian) in both the earnings and selection models. Ethnic variables may be correlated with other characteristics of the individual such as educational attainment, which is problematic. The quantitative estimates of the wage premium and returns to education did not change significantly, which was reassuring. 10. Nonlabor income is the sum of rental income, investment income, remittances, pensions, and other transfers not involving payment in exchange for the provision of labor by the household. 11. I exclude marital status, household composition, and the individual s unearned income from the multinomial logit equations as they should rightfully be treated as endogenous variables. 12. This excludes outliers, individuals whose earnings were more or less than three standard deviations from a normalized mean. I also limited the analysis to Malaysian nationals since non-malaysians cannot be employed in public administration, hence, it would be outside their employment choice set. In addition, I excluded workers in agriculture, forestry, and fisheries as it is typically a sector that few choose to join. 13. All the wages and income are quoted in 2000 MYR at USD 1.00 = MYR In the public sector, only 7.0 percent of males and 2.5 percent of females have completed just primary schooling compared to 15.6 percent of males and 11.0 percent of females in the private sector. 15. Most of the explanatory variables used in the multinomial regression, as shown in Tables 2 and 3, are statistically significant. The low R 2, though not uncommon in the literature for labor participation and wage regressions, does suggest that there are still other (mostly unobserved) variables that could be included to improve the overall fit. However, as long as these unobserved regressors are orthogonal to the included control variables or do not affect their direction, we should not be too concerned about their exclusion. 16. I exclude unearned income of the individuals due to endogeneity concerns that is, employment in the labor force may affect the probability of individuals pursuing opportunities for unearned income. 17. These are the exponentiated result values from the observed wage gap, which are currently expressed in logarithmic scale.

17 November December 2013, Volume 49, Supplement For male workers, the share of the public-private wage differential that is explained by endowments is obtained by dividing the explained difference by the total adjusted difference: 0.165/0.571 = A similar approach is used to compute the share for female workers. 19. Discrimination against women in salaries, after controlling for human capital endowments and personal attributes, was also found in Turkey by Tansel (2005). References Assaad, R The Effect of Public Sector Hiring and Compensation Policies on the Egyptian Labor Market. World Bank Economic Review 11, no. 1: Choong, C.; Mahyudidin, Z.; Ridzwan, D.; and E. Mohd Azrai OLG Macro-Modeling Project: Knowledge, Innovation and Growth: Labor Markets and Innovation in Malaysia. KRIS Views, Khazanah Nasional Berhad. Danh, N., and H. Long Public-Private Wage Differentials for Males and Females in Vietnam. MPRA Paper no. 6738, University Library, Munich. Fernandez, J.L Gender Earnings Differentials in Malaysia. Ph.D. dissertation, University of Malaya, Kuala Lumpur. Heckman, J Sample Selection Bias as a Specification Error. Econometrica 47, no. 1: Hou, J Public-Private Wage Comparison: A Case Study of Taiwan. Journal of Asian Economics 4, no. 2: Lee, L.-F Generalized Econometric Models with Selectivity. Econometrica 51, no. 2: Lindauer, D.L., and R.H. Sabot The Public/Private Wage Differential in a Poor Urban Economy. Journal of Development Economics 12, no. 1: Mazumdar, D The Urban Labor Market and Income Distribution: A Study of Malaysia. New York: Oxford University Press and World Bank. Mincer, J Schooling, Experience, and Earnings. New York: National Bureau of Economic Research. Nielsen, H.S., and M. Rosholm The Public-Private Sector Wage Gap in Zambia in the 1990s: A Quantile Regression Approach. Empirical Economics 26, no. 1: Tansel, A Public-Private Employment Choice, Wage Differentials, and Gender in Turkey. Economic Development and Cultural Change 53, no. 2: Terrell, K Public-Private Wage Differential in Haiti. Journal of Development Economics 42, no. 2: Trost, R., and L.-F. Lee Technical Training and Earnings: A Polytomous Choice Model with Selectivity. Review of Economics and Statistics 66: United Nations Malaysia: Public Administration Country Profile. New York (available at World Bank Malaysia Productivity and Investment Climate Update. East Asia and Pacific Region Report no MY, Washington, DC Malaysia Economic Monitoring: Inclusive Growth. Economic Updates and Modeling Report no , Washington, DC. To order reprints, call ; outside the United States, call

18 Copyright of Emerging Markets Finance & Trade is the property of M.E. Sharpe Inc. and its content may not be copied or ed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or articles for individual use.